As security needs increase, the scale of a video surveillance system is increasing. Further, the definition of video is getting higher, and a period for storing video data is getting longer. Accordingly, the volume of video data to be stored in a video management system is increasing at an accelerating pace. Further, along with automatic detection of persons or objects by analyzing surveillance camera video images, there is also a need to store a large amount of high-quality data used as study data.
In view of such a situation, storage in a local PC is difficult to satisfy a capacity requirement for large-volume video data. Therefore, a method of arranging video data on cloud storage, having an architecture in which a large capacity and high reliability are secured, is considered. It should be noted that cloud storage means a service of lending an external storage region (storage) on the Internet.
Patent Document 1: JP 2001-525636 A
In general, however, cloud storage has the following characteristics unlike direct storage:                As it is provided at a remote site, it is not guaranteed that a network communication band is secured sufficiently, whereby communication delay becomes large; and        There are consistency constraints in order to secure a large capacity and high reliability, so it is not guaranteed that recorded data can be read at any time.        
It should be noted that cloud storage or the like usually has a structure of blocking reading and writing of data until data is reproduced for backup to maintain consistency.
With such characteristics as described above, in the case of placing video data on cloud storage, there is a problem that necessary video data cannot be acquired immediately whereby video data cannot be viewed on the local PC side in real time.
Specifically, the following problems are caused:
(1) When video distribution in an extremely large band is required like a case where video images of a plurality of cameras are viewed simultaneously, video distribution cannot be performed due to a shortage in the network communication band. Even in the case of performing video analysis, large-volume video data must be transmitted to an analysis server, so it takes time to acquire video data;
(2) When attempting to perform special operation (jump, forward, or rewind) at the time of reproducing video images, response to an operation is degraded due to a communication delay between the cloud storage and a local PC (PC on which video images are to be reproduced); and
(3) When attempting to reproduce or analyze video images stored immediately before such as several seconds to several tens of seconds before, the video images may not be acquired immediately due to consistency constraints of the cloud storage.
As a technology of storing video data, there is a technology disclosed in JP 2001-525636 A (Patent Document 1). Patent Document 1 discloses a method of dividing video data and storing the divided pieces of data in different files in order to reduce the load on a server device that distributes the video data. Specifically, in order to reduce the load to carry out high-speed reproduction, compressed data is divided into anchor picture frames (I frames) and predictive picture frames (P frames and B frames), which are stored in different files.
With such a method, however, there is a problem that it is still impossible to reproduce video data stably at a high speed as described above. In other words, in Patent Document 1, the load on the server is reduced only because data required for high-speed reproduction is divided in advance. Therefore, there still are problems such as distribution inability or a delay due to a shortage of network communication band, lowering of response to reproduction operation, and a delay in video capturing due to consistency constraints of cloud storage.
Moreover, there is also a problem that processing cannot be performed stably at a high speed not only in the case of reproducing video data but also in the case of performing any processing on video data such as analysis of video data.